ALARMS WERE DEVELOPED to improve patient safety, but alarm fatigue may actually put patients at higher risk for harm. The literature identifies alarm fatigue as a lack of response to an alarm due to an excessive number of alarms, resulting in sensory overload and desensitization.1 Redundant alarms can interfere with patient care and delay nurse response times.2 The ECRI Institute identified alarm fatigue as a top health hazard from 2012 to 2015.3-6 Cardiac monitoring alarms are among the top contributors to alarm fatigue in most hospitals.
This article follows the journey of one acute care institution's search for a better alarm management solution using new technology to replace its beeper-based system for telemetry alarm events.
Cardiac monitoring and alarm fatigue
Cardiac monitoring provides caregivers with continuous electrocardiography with heart rate tracking, rhythm interpretation, and audible alarms for identified conditions. Telemetry monitoring is a form of cardiac monitoring where each patient wears a wireless transmitter that sends continuous electrocardiography to the central station. This primary monitor is where all real-time tracings of the heart's electrical activity are analyzed and recorded, and where alarms are generated.
In 2017, the American Heart Association (AHA) published the updated practice standards for the use of cardiac monitoring in the hospital setting. The classification of recommendations and level of evidence in the practice standards define Class I (monitoring always required), Class IIa and IIb (monitoring may be required), Class III (no benefit), or Class III (harmful).7 Cardiac monitoring in non-ICU settings has increased, and patients are often monitored for reasons not supported by AHA practice standards.8-11
Recognizing the dangers inherent in clinical alarms and the potential for serious patient harm, The Joint Commission put forth a national patient safety goal. By 2014, hospitals were required to establish alarm system safety as a priority and identify strategies to improve the safety of clinical alarm systems. By 2016, they were required to establish policies and procedures for managing alarms and educate staff about the purpose and proper operation of alarm systems for which they are responsible.12
An analysis of the telemetry monitoring needs at one 473-bed acute care institution revealed the need to replace the notification device for telemetry alarm events. At this institution, five inpatient units have telemetry monitoring capabilities, and the total number of telemetry beds at the time of this project was 74. Each patient on telemetry monitoring wears a wireless telemetry box that sends the activity of the heart to the central station. In select clinical areas, the patient is also monitored with a bedside monitor. To allow for timely alarm response, support patient safety, and optimize nursing
workflow, nurses use an alarm notification device on the five units with telemetry monitoring capabilities. The institution used a beeper-based system prior to the start of this project that notified assigned nurses of all crisis and warning level alarms. Unanswered alerts were forwarded to all nurses (telemetry technicians are not used).
Problems with the existing beeper-based system included the inability to obtain replacement parts and system maintenance from the vendor. The system did not provide alarm filtering, so the primary monitor sent every crisis and warning level alarm to the beeper device. All unit alarms were sent in one queue, meaning only one alarm was transmitted at a time, resulting in delays during times of high alarm volume. Staff reported alarm fatigue and a lack of confidence in the system. An opportunity existed to use new technology to meet the requirements of The Joint Commission's safety goal, support best practices in alarm management, and reduce alarm redundancy.
In response to The Joint Commission's safety goal, this institution created an Alarm Management Committee composed of physicians, nurses, administrators, and technical staff. A primary assignment was identifying and selecting a replacement alarm notification device. The new device needs to support the requirements of the safety goal, provide for alarm filtering, reduce staff alarm fatigue, and reflect current best clinical practices for managing alarms.
The committee interviewed multiple vendors; deliverables included sending the alarm to an endpoint device with patient name, room number, and an event waveform. No one vendor could provide all required elements and most offered less functionality than the current device. The committee discovered a smartphone alarm management application (app), waveform app, and middleware software system used together at another institution that could meet all the goals. Middleware software connects client-based network requests to the data needed.13 In this example, the software bridges the telemetry system with the smartphone app. The committee selected this system as the solution to replace the beeper system.
From this larger organizational committee, a core multidisciplinary project team was created with representatives from information technology, nursing informatics, and clinical nursing departments. Using Plan-Do-Study-Act performance improvement methodology, the team worked collaboratively with vendors to implement the new smartphone solution. The project team provided periodic updates to the alarm management committee.
A collaborative working relationship with the four vendors was key to the project's success. Vendors participated in weekly meetings to review project deliverables, status, and issues. Technical deliverables included infrastructure readiness, multisystem integrations, and information security reviews. To engage the clinical nurses and solicit feedback on current alarm management, each vendor did walking rounds on each monitoring unit. Most nurses mentioned alarm fatigue and the tachycardia alarm volume as most bothersome. A prepilot survey was conducted and analyzed to further explore nurses' perception of alarms and alarm fatigue. Participation in the survey was voluntary.
A nursing superuser group, with representation from each unit, met weekly with the clinical nurse specialists, nursing professional development specialists, and nursing informatics staff. The superuser group supported the core project team's efforts by agreeing on key components of the nursing policy and developing safe workflows with the new device. For example, the new system automatically sends a sample alert to the smartphone every 4 hours. This alert confirms nurse assignments for each telemetry patient and reminds nurses to return to the primary system and/or central station to review alarm history every 4 hours. Charge nurse responsibilities, including verifying the accuracy and equity of the telemetry assignment and creating the assignment in the middleware software system, were delineated. Superuser responsibilities were standardized across the institution; superusers shared project status updates and meeting minutes with their colleagues to facilitate nursing engagement.
The four components include the telemetry monitoring system, middleware software, smartphone alarm app, and waveform app. As the primary monitoring system, the telemetry system analyzes patient cardiac rhythms and generates alarms. The middleware software receives alarm information from the primary system and forwards select filtered alarms to the smartphone. It also records and reports a time stamp for alarm receipt and acknowledgment. The alarm management smartphone app provides ringtone alerts and banner notifications with patient alarm information including patient name, alarm type, time, and room number. Nurses acknowledge the alarm condition and launch the waveform app with a single swipe on the screen. The mobile waveform app allows nurses to view both the corresponding static alarm waveform or event strip and livestream waveforms of the patient's current rhythm at the point of care. Nurses receive alarms for their assigned patients. Select alarms escalate to all unit staff if not responded to within a designated amount of time. The team selected a smartphone device that could house these two apps, meet future clinical needs for bar code scanning and secure text messaging, and was durable in the clinical environment.
The new system allowed the project team to access and analyze alarm data, which helped identify and confirm alarm priorities, as well as frequency and length of alarm types. Tachycardia was identified as an outlier alarm occurring frequently on all telemetry units. This supported anecdotal reports from nurses of a fatiguing amount of tachycardia alarms. Baseline alarm data from the pilot unit during a 5-day period showed 1,972 dysrhythmia alarms, of which 1,771 were tachycardia. The team analyzed the average length of tachycardia alarms and noted most alarms were less than 10 seconds and attributable to a small number of outlier patients.
At this institution, the tachycardia parameter is set at 140 beats/minute as the high heart rate parameter. When a patient alarms for tachycardia, nurses collaborate with the medical team to treat the underlying condition, escalate care to the rapid response team if needed, and customize the alarm parameters to reduce redundant alarms. Despite these interventions, tachycardia continued to alarm frequently.
Based on data, patient population, and trial-and-error testing, all tachycardia events that were sustained for at least 10 seconds were sent to the smartphone. This 10-second delay reduced the number of short duration tachycardia events sent to the smartphone. Such events reflected clinically insignificant brief heart rate elevations and artifact alarms. A tachycardia count trigger alarm was created to manage tachycardia events that were shorter in duration and occurred frequently. This alarm counts and captures six tachycardia events that are less than 10 seconds in length and occur within 2 minutes. The six tachycardia alarms are bundled into one alert for the nurses on their smartphones, potentially reducing alarm fatigue. (See Count trigger tachycardia alerts.)
Filters were created for other alarm conditions. For example, ventricular couplets labeled as VT>2 (ventricular tachycardia greater than two) alarms are bundled and sent as one count trigger alarm when they occur frequently. Delays were placed on system alarms. On units with pulse oximetry monitoring, filters were created to reduce alarm redundancy.
Multiple design features were added to improve design, functionality, and patient safety. The ringtones on the alarm app were set to match the same audible alarms from the primary central station so staff did not have to relearn and manage new sounds. The slide button to acknowledge an alarm was customized to automatically launch the waveform app, display the event strip, and capture clinical strip acknowledgment. Multiple rounds of testing and troubleshooting by the project team and vendors improved workflow, alarm configuration safety, and secondary alarm management by clinical staff.
To prepare for implementation, training sessions for nurses were collaboratively created and presented by the clinical nurse specialist and the vendor technical trainers. Training provided an overview of the secondary alarm management system and included hands-on training of new smartphones and alarm simulations. Workflow expectations for securing and inventorying the smartphones, cleaning the smartphones between use, and troubleshooting both the middleware assignment and smartphone app were reviewed. Over 300 nursing and nonnursing employees attended the training sessions. Additional simulation-based staff development programs during implementation contributed to nursing skill and confidence.
The pilot was conducted on one telemetry unit over a 6-week period before expansion to the remaining units. The pilot allowed the team to fine-tune the alarm filters to support safe patient care and better reflect nursing workflows. Daily issue calls allowed for an organized team-oriented approach to solving problems. For example, the team increased the time the nurse is allotted to acknowledge an alarm before it escalates. Based on nursing feedback, the smartphone ringtones were lengthened to ensure they were easily heard. The project team and vendors provided onsite support for the first week of the pilot. The project was then implemented on the five remaining units over 3 months.
Using the middleware vendor application to analyze data, the project team examined alarm data from the pilot unit. During a 1-week time frame, patients alarmed a total of 6,884 times for telemetry alarms inclusive of all crisis, warning, advisory, and message alarms recorded at the primary alarm with tachycardia, representing 59% of the alarms, or 4,048 alarms. Of the total 4,048 tachycardia alarms, 440 alarms were longer than 10 seconds and sent immediately to the smartphone. Of the total 4,048 tachycardia alarms, 2,610 alarms were less than 10 seconds in length, but occurred frequently in a short period of time (six times in 2 minutes), meeting the system's requirement to notify nursing. These alarms were bundled into 435 alarms labeled as “tachycardia count trigger” alarms for the smartphone. Of the total 4,048 tachycardia alarms, 998 were less than 10 seconds, did not occur frequently, did not meet the filtering criteria, and only alarmed at the primary monitor or central station. The nurses viewed these alarms at the primary monitor every 4 hours and as needed during alarm reconciliation.
Seventy-five percent, or 3,050, of the original 4,048 tachycardia alarms were filtered and bundled into a total of 875 tachycardia and tachycardia count trigger alarms and delivered to nurses' smartphones, reducing the total number of alarms by 78% when considering the institution's former beeper-based system would have delivered all 4,048 alarms.
To better understand nurses' perceptions of alarm fatigue and incorporate nursing feedback into the design of the secondary alarm device, a pre- and postsurvey of nurses was conducted on each unit using an adapted version of the 2011 National Health Technology Foundations Clinical Alarms Survey.14 The survey was distributed via e-mail to RNs working on each of the monitoring units, and participation was voluntary.
The sample size reflects the number of nurses who completed the survey. See Survey results (pre- and postsurveys). Postsurvey data showed a 16.2% reduction to “agree/strongly agree” responses that nuisance alarms occur frequently, and a 16.4% increase to “agree/strongly agree” responses that alarms are adequate to alert staff of potential or actual changes in patient conditions.
Using middleware technology to filter and bundle the most clinically significant alarms to the nurse provides real-time alerts and escalations for urgent alarms while also greatly reducing the number of redundant alarms. The success of alarm filtering and bundling is reflected in the 78% fewer tachycardia alarms received by nurses on the pilot unit. Nurses continue to reconcile all alarms at the Central Station every 4 hours and as needed throughout the shift. No adverse patient events have been noted. The app allows for clear and concise alarm recognition and reconciliation. Acknowledged versus unacknowledged alarms are different colors and fonts so it is more difficult for nurses to miss an alarm. During times of high alarm volume, alarms are efficiently sent to the smartphone without delays related to alarm traffic. This is a significant improvement compared with the beeper-based system. Having both the static and livestreaming waveforms on the smartphone allows the nurse to access vital clinical information at the point of care, respond to changes in patient condition quickly, and maximize nursing workflow efficiency.
Upon admission to a telemetry bed, patients, caregivers, and visitors are educated about the use of smartphones for alarm management. Anecdotal patient and caregiver feedback has been positive.
Engaging nurses throughout the process was integral to fostering engagement and enthusiasm during the rollouts. Incorporating nurses' feedback into the design of the smartphone increased user satisfaction. Adjusting alarm filtering and escalation times based on nursing feedback during the pilot allowed for safe and efficient expectations in alarm response. Nurses' perceptions of enhanced confidence in alarms are reflected in the pre- to postsurvey results.
The project team and vendor partnership was successful because of clearly delineating responsibilities, timelines for completion of tasks, and appropriate allocation of human resources to complete required tasks.
This organization is working toward leveraging available alarm data into meaningful reports. The organizational alarm management committee is currently exploring reporting options that illustrate alarm response timeliness and examine individual nurse alarm burden.
The smartphone device is being piloted for additional alarm management purposes, such as nurse call and sepsis alerts. Future expansions for more communication functionality, such as Voice over Internet Protocol calling and Health Insurance Portability and Accountability Act-compliant secured text messaging, are being explored. The smartphone is equipped with a bar code scanner, and other project teams are working to test, plan, and support knowledge-based bar code scanning for medication administration.
Customizing smartphone and middleware technology revolutionized management of clinical alarms in real time and at the point of care. This model can serve other organizations in their efforts to standardize alarm configuration and escalation pathways, manage a highly complex technical project, and successfully implement a new system across multiple monitoring units through onsite education and staff engagement.